2 * Copyright (c) 2015 The WebM project authors. All Rights Reserved.
4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
11 #ifndef VPX_VPX_DSP_MIPS_TXFM_MACROS_MSA_H_
12 #define VPX_VPX_DSP_MIPS_TXFM_MACROS_MSA_H_
14 #include "vpx_dsp/mips/macros_msa.h"
16 #define DOTP_CONST_PAIR(reg0, reg1, cnst0, cnst1, out0, out1) \
18 v4i32 s0_m, s1_m, s2_m, s3_m, s4_m, s5_m; \
19 v8i16 k0_m, k1_m, k2_m, zero = { 0 }; \
21 k0_m = __msa_fill_h(cnst0); \
22 k1_m = __msa_fill_h(cnst1); \
23 k2_m = __msa_ilvev_h((v8i16)k1_m, k0_m); \
24 k0_m = __msa_ilvev_h((v8i16)zero, k0_m); \
25 k1_m = __msa_ilvev_h(k1_m, (v8i16)zero); \
27 ILVRL_H2_SW(reg1, reg0, s5_m, s4_m); \
28 ILVRL_H2_SW(reg0, reg1, s3_m, s2_m); \
29 DOTP_SH2_SW(s5_m, s4_m, k0_m, k0_m, s1_m, s0_m); \
30 s1_m = __msa_dpsub_s_w(s1_m, (v8i16)s5_m, k1_m); \
31 s0_m = __msa_dpsub_s_w(s0_m, (v8i16)s4_m, k1_m); \
32 SRARI_W2_SW(s1_m, s0_m, DCT_CONST_BITS); \
33 out0 = __msa_pckev_h((v8i16)s0_m, (v8i16)s1_m); \
35 DOTP_SH2_SW(s3_m, s2_m, k2_m, k2_m, s1_m, s0_m); \
36 SRARI_W2_SW(s1_m, s0_m, DCT_CONST_BITS); \
37 out1 = __msa_pckev_h((v8i16)s0_m, (v8i16)s1_m); \
40 #define DOT_ADD_SUB_SRARI_PCK(in0, in1, in2, in3, in4, in5, in6, in7, dst0, \
43 v4i32 tp0_m, tp1_m, tp2_m, tp3_m, tp4_m; \
44 v4i32 tp5_m, tp6_m, tp7_m, tp8_m, tp9_m; \
46 DOTP_SH4_SW(in0, in1, in0, in1, in4, in4, in5, in5, tp0_m, tp2_m, tp3_m, \
48 DOTP_SH4_SW(in2, in3, in2, in3, in6, in6, in7, in7, tp5_m, tp6_m, tp7_m, \
50 BUTTERFLY_4(tp0_m, tp3_m, tp7_m, tp5_m, tp1_m, tp9_m, tp7_m, tp5_m); \
51 BUTTERFLY_4(tp2_m, tp4_m, tp8_m, tp6_m, tp3_m, tp0_m, tp4_m, tp2_m); \
52 SRARI_W4_SW(tp1_m, tp9_m, tp7_m, tp5_m, DCT_CONST_BITS); \
53 SRARI_W4_SW(tp3_m, tp0_m, tp4_m, tp2_m, DCT_CONST_BITS); \
54 PCKEV_H4_SH(tp1_m, tp3_m, tp9_m, tp0_m, tp7_m, tp4_m, tp5_m, tp2_m, dst0, \
58 #define DOT_SHIFT_RIGHT_PCK_H(in0, in1, in2) \
63 DOTP_SH2_SW(in0, in1, in2, in2, tp1_m, tp0_m); \
64 SRARI_W2_SW(tp1_m, tp0_m, DCT_CONST_BITS); \
65 dst_m = __msa_pckev_h((v8i16)tp1_m, (v8i16)tp0_m); \
70 #define MADD_SHORT(m0, m1, c0, c1, res0, res1) \
72 v4i32 madd0_m, madd1_m, madd2_m, madd3_m; \
73 v8i16 madd_s0_m, madd_s1_m; \
75 ILVRL_H2_SH(m1, m0, madd_s0_m, madd_s1_m); \
76 DOTP_SH4_SW(madd_s0_m, madd_s1_m, madd_s0_m, madd_s1_m, c0, c0, c1, c1, \
77 madd0_m, madd1_m, madd2_m, madd3_m); \
78 SRARI_W4_SW(madd0_m, madd1_m, madd2_m, madd3_m, DCT_CONST_BITS); \
79 PCKEV_H2_SH(madd1_m, madd0_m, madd3_m, madd2_m, res0, res1); \
82 #define MADD_BF(inp0, inp1, inp2, inp3, cst0, cst1, cst2, cst3, out0, out1, \
85 v8i16 madd_s0_m, madd_s1_m, madd_s2_m, madd_s3_m; \
86 v4i32 tmp0_m, tmp1_m, tmp2_m, tmp3_m, m4_m, m5_m; \
88 ILVRL_H2_SH(inp1, inp0, madd_s0_m, madd_s1_m); \
89 ILVRL_H2_SH(inp3, inp2, madd_s2_m, madd_s3_m); \
90 DOTP_SH4_SW(madd_s0_m, madd_s1_m, madd_s2_m, madd_s3_m, cst0, cst0, cst2, \
91 cst2, tmp0_m, tmp1_m, tmp2_m, tmp3_m); \
92 BUTTERFLY_4(tmp0_m, tmp1_m, tmp3_m, tmp2_m, m4_m, m5_m, tmp3_m, tmp2_m); \
93 SRARI_W4_SW(m4_m, m5_m, tmp2_m, tmp3_m, DCT_CONST_BITS); \
94 PCKEV_H2_SH(m5_m, m4_m, tmp3_m, tmp2_m, out0, out1); \
95 DOTP_SH4_SW(madd_s0_m, madd_s1_m, madd_s2_m, madd_s3_m, cst1, cst1, cst3, \
96 cst3, tmp0_m, tmp1_m, tmp2_m, tmp3_m); \
97 BUTTERFLY_4(tmp0_m, tmp1_m, tmp3_m, tmp2_m, m4_m, m5_m, tmp3_m, tmp2_m); \
98 SRARI_W4_SW(m4_m, m5_m, tmp2_m, tmp3_m, DCT_CONST_BITS); \
99 PCKEV_H2_SH(m5_m, m4_m, tmp3_m, tmp2_m, out2, out3); \
101 #endif // VPX_VPX_DSP_MIPS_TXFM_MACROS_MSA_H_